CAIN-42 CAIN StudioCEFV · Module 5 of 5 · 11 min
Grading facts: observed, verified, derived, unverified, counterfactual
Every statement in an evidence report belongs to a layer. OBSERVED: seen directly. VERIFIED: re-checked independently. DERIVED: computed from other facts. UNVERIFIED: claimed, not checked. COUNTERFACTUAL: what would have happened otherwise.
Mixing layers is how reports mislead: a counterfactual 'would have blocked' presented as an observed block; a founder's claim presented as verified revenue.
Label every line, and never promote a statement without new evidence.
Key points
- Five layers; label every statement.
- Never promote without new evidence.
- Counterfactuals stay counterfactual.
In the CAIN-42 codebase
cain/passport.py— TrajectoryPassport 5-layer fact segregation
Practice check
Practice 1 · choose one
'Our engine would have blocked this attack' is…
Practice 2 · true or false